Plots: 500 observations

Bias of estimates

1st version Remark: I have inserted the error bar associated to the bias but most of the times the distribution is mostly centered around the mean so is not visible

Bias of estimates

2nd version Pro: unique plot, easier to compare trends between different observation schemes Cons: sometimes more difficult to highlight trend of different models in same obs scheme I am gonna keep this for next plots

Coverage of estimates

Error type I in different models

1st version Pro: more aesthetic Cons: difficult to detect which dots belong to which model

Error type I in different models

2nd version Pro: easier to interpret and intuitive Cons:I am not convinced by the bars I am gonna keep this version for next plots

Power of different models

Distribution of estimands associated to covariates

Relative Bias of time spent in each state

Plots: 2K observations

Bias of estimates

Coverage of estimates

Error type I in different models

Power of different models

Distribution of estimands associated to covariates

Relative Bias of time spent in each state

Plots: 5K observations

Bias of estimates

Coverage of estimates

Error type I in different models

Power of different models

###Distribution of estimands associated to covariates

Relative Bias of time spent in each state

Plots: 10K observations

Bias of estimates

Coverage of estimates

Error type I in different models

Power of different models

Distribution of estimands associated to covariates

Relative Bias of time spent in each state

Comparing performances for different samples sizes

Convergence of different models

Averaged computational time for different models

Bias of estimates comparison

Coverage comparison (1 year)

Coverage comparison (3-6 years)

Error type I comparison

Power comparison

Comparison coefficient distribution (3-6 years)

Relative Bias of time spent in each state (3-6 years obs scheme)